Performance Implications of Hyper-Threading

1 Ethan Bolker is also Professor of Computer Science at the University of Massachusetts, Boston. Intel’s recently introduced Hyper-Threading Technology (HTT) makes a single physical processor appear as two logical processors. Operating systems and applications can schedule processes or threads on those logical processors. The performance impact of HTT varies, depending on the nature of the applications running on the processors and on how the hardware is configured. This paper illustrates that variation by presenting a set of controlled studies and offers a simple queueing model to interpret the observations. We also examine the issues involved in collecting performance metrics for HTT systems and their implications for modeling.